Volume Completion Between Contour Fragments at Discrete Depths

Area of Science:

• Visual Perception
• Computational Neuroscience
• 3D Surface Reconstruction

Background:

• Illusory contours and surfaces are well-studied phenomena in visual perception.
• Previous work demonstrated 3D curved, nonclosed illusory surfaces from binocular stereo viewing.
• Existing models struggle to explain complex 3D surface perception from limited cues.

Purpose of the Study:

• To investigate the perception of closed curved surfaces and volumes from binocularly fused images.
• To explore how the visual system interpolates surfaces between discrete depth layers.
• To constrain models of 3D surface and volume generation in visual processing.

Main Methods:

• Utilizing binocular stereo viewing of disparate images with discrete contour elements.
• Analyzing the percept of closed curved surfaces and volumes generated by these stimuli.
• Examining the role of contour continuation and inferred 'rim' segments in surface interpolation.

Main Results:

• Binocular presentation elicits percepts of closed curved surfaces/volumes across discrete depths.
• Surfaces are filled in via virtual contours interpolated from visible contour segments.
• Illusory surfaces exhibit consistent depth and curvature despite absent local disparity cues.

Conclusions:

• The brain constructs coherent 3D surfaces and volumes from sparse binocular disparity information.
• Perception of smooth, closed surfaces relies on contour continuation and inferred surface geometry.
• These findings challenge current visual processing models and inform theories of 3D scene reconstruction.